AMI Labs, World Labs Won't Say What Their World Models Will Sell

AMI Labs and World Labs, the two best-funded companies building so-called world models, will not say which commercial products they intend to ship or when, according to a TechCrunch account of a panel at the All In conference this week, first reported by TechCrunch.

What is a world model, and what does it bind?

A world model is a system trained to predict how physical space behaves — the same underlying approach that lets a self-driving car anticipate traffic, according to the TechCrunch report. The simplest version functions as a navigable map of the world. That same architecture, per the report, can be pointed at a humanoid robot learning to carry boxes, or at raw video footage that a model turns into an explorable 3D environment. The technology does not bind a single industry the way a piece of legislation binds a regulated sector; it binds whichever product team decides to build on top of it, which is precisely why outsiders cannot yet say where the money will land.

What are AMI Labs and World Labs actually saying?

Michael Rabbat, a co-founder of AMI Labs and the company's VP of World Models, appeared on the panel and was pressed on the company's plans. His answer, as quoted by TechCrunch: "We'll talk about it when we're ready to talk about it." Asked to clarify by email, Rabbat told TechCrunch: "We're still in a research and building phase, so we're not talking publicly about any product plans or timeline."

World Labs, co-founded by Fei-Fei Li, has gone further toward a public product than AMI has. Its tool, Marble, is described in the TechCrunch report as the most fully developed offering in the space, with demonstrations spanning media creation, explorable game environments, and CGI effects, along with robotics use cases. Even so, the report characterizes the platform as designed primarily to demonstrate capability rather than to signal a settled commercial direction.

Why won't the companies name a target market?

AMI Labs, founded less than a year ago according to TechCrunch, has already touched manufacturing, biomedicine, robotics, and clinician-facing software through a partnership with Nabia. TechCrunch's framing is direct: the company is unlikely to pursue all of those markets at once, but nothing in its public statements indicates which one or two will win out.

The report attributes the silence partly to funding conditions rather than product uncertainty alone. As long as capital keeps flowing to world-model research without demanding a near-term product roadmap, TechCrunch notes, there is no financial pressure to commit to one vertical. The report also flags a strategic reason to stay quiet: announcing a specific application — a humanoid robot platform, a Hollywood production tool — would tip competitors toward the same target before a company has locked in an advantage.

How does the secrecy reach outside these companies?

The opacity is not confined to the labs building the models. Alex de Vigan, CEO of Physicl, a data supplier to world-model companies, told TechCrunch on the sidelines of the same conference that he does not know what his own data is being used to build. "I wish they would tell us more," de Vigan said, according to the report. "We could build more useful data if we knew what they were working on." That comment establishes that the information gap runs through the supply chain, not just between founders and the public — a vendor providing input data cannot optimize what it sells without knowing the end use.

What to watch

• Whether AMI Labs makes any public statement narrowing its focus beyond the manufacturing, biomedicine, robotics, and Nabia clinician-software work already disclosed.
• Whether World Labs expands Marble beyond media and game-environment demos into a named robotics or self-driving product.
• Whether data suppliers such as Physicl gain more visibility into end-use cases, which de Vigan says would let them build more targeted datasets.
• Whether funding conditions for world-model startups tighten, which — per the dynamic TechCrunch describes — would be the likeliest trigger for a company to commit publicly to one commercial vertical.

No party quoted in the TechCrunch report disputes that multiple viable businesses could be built on world-model technology. The disagreement, such as it is, sits between founders who see no reason to specify one path while capital remains easy to raise, and outside observers — including at least one of their own suppliers — who say more disclosure would make the technology easier to build around.